From C# we have to call python function, in this case the nltk lib lemmatize function (which still has no good C# implementation). We call it like this:

private string Lemmatize(string word)
{
 using (Py.GIL())
 {
   using (var scope = Py.CreateScope())
   {
     dynamic nltk = Py.Import("nltk");
     var lemmatizer = nltk.stem.WordNetLemmatizer();
     var lemma = lemmatizer.lemmatize(word);
     return lemma;
  }
 }
}

It works perfectly, in the test application, from a .net console app. When we apply this into the real app, which has async functions we experienced that the function usually hang at the point using (Py.GIL()). The whole app runs single threaded, because async functions everywhere, but each of them is called with await. Therefore a calling point looks like this:

private async Task<IList<string>> LemmatizeAll(IEnumerabe<string> words)
{
   var result = new List<string>();
   foreach(var word in words)
   {
      var item = Lemmatize(word);
      result.Add(item);
   }
   return result;
}

//

var lemmas = await LemmatizeAll(new [] {"running", "man"});

I understand that the Py.GIL() is mostly a semaphore, not enabling running multiple threads at the same time, but this not what happens now. It runs from only one thread, however from different threads because of the async-await. To be hung is not a nice thing from it.

How to use the python scripts from an environment like this? Any advice?